中国商业银行信用风险度量研究

信用风险是商业银行面临的主要风险,信用风险的度量模型有专家判断法、信用评分法、神经网络分析法以及现代违约概率模型等。通过比较分析LOGIT模型和KMV模型,选取了能够体现公司盈利能力、营运能力、资本结构、偿债能力、成长能力和现金流量的28个指标,运用逐步回归方法建立LOGIT模型,发现该模型能够提前一年较好地预测出公司的违约情况。在分析KMV模型时,通过GARCH-M模型计算出企业股权价值波动率,并运用上市公司数据得出样本公司的股权价值和违约点,从而计算出样本公司的资产价值和资产价值波动率,最后得出KMV模型的判别结果。上述分析表明我国商业银行应以LOGIT模型作为判别模型,以KMV模型作为追踪模型,将LOGIT模型与KMV模型相结合来判断贷款企业的信用风险水平。

A Study on Credit Risk Measurement of Chinese Commercial Banks

Credit risk is the main risk taken by commercial banks.Credit risk measurement models include Expert Judgment,Credit Scoring,Neural Network Analysis as well as Modern Default Probability Model.In this paper,LOGIT model and KMV model are compared.The 28 indexes are selected to reflect the company＇s profitability,operating capabilities,capital structure,solvency,growth ability and cash flows.Forward Stepwise Regression is used to establish LOGIT model,which can predict company default 1 year before actual default.Then Assets Values of KMV model is calculated,through GARCH-M model Asset Volatility is calculated,and then the default rates of sample companies are estimated.Finally we get the discriminating result.Through comparison we bring forward that LOGIT model is fit for discriminating,at the same time KMV model is better for tracing.These two models should be combining to judge credit risk level of corporation in debt.